As AI systems evolve beyond passive prediction toward autonomous decision-making and proactive interaction, their growing capacity to act in complex environments introduces new scientific and societal challenges. This shift toward agentic AI raises fundamental questions about design principles, alignment, and oversight, as well as their impact on digital ecosystems, especially as scientific progress accelerates and adoption scales globally.

These developments call for cooperation between research, industry, and policy communities for better mutual understanding, and to ensure that agentic AI is deployed responsibly and effectively. The goal of this symposium is to foster an accessible, research-driven dialogue exploring both scientific foundations and practical implications for industry and public policy.

Held in anticipation of the international India Impact Summit, this event will showcase key initiatives from Criteo and leading organizations. Across three accessible sessions covering the fundamentals and the various faces of agentic AI, its impact on the open Internet and digital economy, and its scientific underpinnings, we aim to shape a shared, multi-disciplinary vision for an innovative, responsible, and impactful agentic AI ecosystem.

The event will be moderated by Liva Ralaivola – VP Research at Criteo and Head of the Criteo AI Lab.

8:45 – 9:30 | Welcome / coffee

9:30 – 9:40 | Opening by Diarmuid Gill – Chief Technology Officer at Criteo

9:40 – 12:20 | 1st session – The fundamentals and various faces of Agentic AI

• 9:40 – 10:10 | Ludovic Denoyer – Scientific Director at IMEC: Agentic AI: A revolution in software engineering

• 10:10 – 10:30 | Éric Bothorel – French MP, Député de la 5^e circonscription des Côtes-d’Armor: IA : répondre de ses actes comme condition de notre liberté d’innover, règlementer pour responsabiliser (session in French)

• 10:30 – 10:45 | Coffee

• 10:45 – 11:50 | Panel: Which path to responsible agentic AI?

  • Michel Combot – Chief Technology, Innovation and AI Officer at CNIL

  • Pagona Tsormpatzoudi – SVP & Assistant General Counsel, Privacy, Data Protection & AI Governance, at Mastercard

  • Martin Signoux – AI policy lead Europe at OpenAI

  • Alice Schoenauer Sebag – AI Safety Staff member at Cohere

  • Panel moderated by Nathalie Laneret – VP Government Affairs & Public Policy at Criteo

• 11:50 – 12:20 | Carina Prunkl – Researcher at Inria: Key Updates on AI Safety

12:20 – 13:40 | Lunch

13:40 – 15:30 | 2nd session – Agentic AI in digital ecosystems

• 13:40 – 14:00 | Vanina Paoli-Gagin – French MP, Sénateur de l’Aube (session in French)

• 14:00 – 14:30 | Patrick Pérez – Chief Executive Officer at Kyutai: Entering the era of world models and physical AI

• 14:30 – 15:30 | Agents for e-commerce

  • Lionel Basdevant – Senior Product Director at Criteo

  • Alberto Lumbreras – Staff Researcher at Criteo

15:30 – 16:00 | Coffee

16:00 – 17:30 | 3rd session – Science of Agentic AI

• 16:00-16:30 | Patrick Gallinari – Professor at Sorbonne Université & Distinguished Researcher at Criteo: Generative foundation models for search and recommendation

• 16:30-17:00 | Maria Teleńczuk – Senior Researcher at Owkin: Stop cleaning data, focus on science: Multi-Agent System for End-to-End Biological Data Processing

• 17:00-17:30 | Grégoire Mialon – Staff Research Scientist at Meta Superintelligence Lab: Evaluating frontier agents

17:30 – 18:00 | Closing: Fireside chat with Anne Bouverot – France’s Special Envoy for AI, Chairperson of Ecole Normale Supérieure, Co-Chair of France’s AI & Digital Council & Liva Ralaivola – VP Research at Criteo and Head of the Criteo AI Lab

18:00 – 21:00 | Cocktails & demos

All sessions, unless specified otherwise, will be held in English.

• Ludovic Denoyer – Scientific Director at IMEC

  • Speaker (1st session)

  • Talk title: Agentic AI: A revolution in software engineering

  • Abstract: The traditional paradigm of software engineering is undergoing a fundamental shift, moving from rigid, deterministic code toward autonomous, goal-oriented systems. This talk explores how the evolution of AI has systematically dismantled the barriers of classical programming through three pivotal transitions: data-driven programming, open-ended models, and natural language as a data structure.

  • Bio: Ludovic Denoyer is a specialist in deep reinforcement learning and natural language processing. He gained international recognition for his pioneering work on unsupervised machine translation and “budgeted” neural networks—architectures designed to be efficient in terms of time and memory. After leading the Agent Research Team at H Company and holding a position as a Research Scientist at Meta AI (FAIR), and Staff Research Scientist at Ubisoft, he is now Scientific Director at IMEC France, in charge of building a fundamental AI research team in Paris.

• Éric Bothorel – French MP, Député de la 5^e circonscription des Côtes-d’Armor

  • Speaker (1st session)

  • Talk title: IA : répondre de ses actes comme condition de notre liberté d’innover, règlementer pour responsabiliser

  • Bio: Éric Bothorel is a Member of the French National Assembly. Before entering public office, he worked in the IT and digital services sector, where he held senior management roles, including Sales Director and Quality Director, within companies specializing in IT distribution and services, notably Infopoint and General Electric. His background combines expertise in digital technologies, service quality, and business development.

• Carina Prunkl – Researcher at Inria

  • Speaker (1st session)

  • Title: Key Updates on AI Safety

  • Abstract: General-purpose AI is improving fast – but progress is uneven, and risk management is struggling to keep pace. Drawing on the two Key Updates of the 2026 International AI Safety Report, this talk synthesises key developments since January 2025 in capabilities, emerging and observed risks, and the evolving landscape of evaluations and safeguards. I focus on what has changed most in the underlying evidence, both in the performance of leading systems and in the documented patterns of misuse and impact, and on how developers’ and policymakers’ toolkits are evolving in response. The aim is to give a clear snapshot of what we know, what we don’t, and what this means for assessing and governing general-purpose AI systems.

  • Bio: Carina Prunkl is a Research Scientist (ISFP) at Inria’s project REGALIA and Lead Writer of the 2026 International AI Safety Report. She is furthermore a Senior Research Fellow at the University of Oxford’s Institute for Ethics in AI and Co-PI on CHAILD, a UKRI funded project on children’s autonomy and AI. Her research focuses on agency, autonomy, and meaningful human oversight, as well as on effective evaluation and governance. She has advised governments and international organisations, including the UK government, UNDP, UNICRI, and the Mexican Senate. She holds a DPhil in Philosophy and an MSt in Philosophy of Physics from the University of Oxford, as well as a Master’s degree in Physics from Freie Universität Berlin.

• Michel Combot – Chief Technology, Innovation and AI Officer at CNIL

  • Panelist

  • Bio: Graduate of the École Polytechnique and Télécom Paris, Michel Combot has 27 years of experience in the digital economy, telecoms and media. Michel Combot is Chief Technology, Innovation and AI Officer at CNIL since October 1st, 2024; prior to this, he was General Delegate of Numeum, an organization representing the digital ecosystem in France, and a member of the board of Digital Europe, its European counterpart.

    For 7 years, he was also Managing Director of the French Federation of Telecoms (FFTélécoms) and permanent delegate of the “digital infrastructures” strategic committee.

    Michel Combot has devoted almost 18 years of his career to the public sector. He has worked for independent administrative authorities, notably as Deputy Director General of ARCEP (Autorité de régulation des communications électroniques, des postes et de la distribution de la presse), then the CSA (Conseil supérieur de l’audiovisuel, now the ARCOM), and in several ministerial cabinets. He has helped define and implement numerous public policies relating to the digital economy, the telecoms industry, media and digital services.

• Pagona Tsormpatzoudi – SVP & Assistant General Counsel, Privacy, Data Protection & AI Governance, at Mastercard

  • Panelist

  • Bio: TBD

• Martin Signoux – AI policy lead Europe at OpenAI

  • Panelist

  • Bio: Martin Signoux is the AI Policy Lead for Europe at OpenAI, where he works with governments, regulators, and civil society to advance AI governance and unlock the economic and social value of frontier technologies.

    A graduate of the School of Public Affairs at Sciences Po and the School of International Relations at Fudan University, Martin has built his career at the intersection of technology and public policy.

    Before joining OpenAI, he served as a Public Policy Manager at Meta, focusing on artificial intelligence and emerging technologies.

• Alice Schoenauer Sebag – AI Safety Staff member at Cohere

  • Panelist

  • Bio: TBD

• Vanina Paoli-Gagin – French MP, Sénateur de l’Aube

  • Speaker (2nd session)

  • Bio: Vanina Paoli-Gagin is a French Senator. She holds a PhD in securities law and a U.S. Master of Laws, and is a qualified lawyer. Her professional background includes work within local government and the French Senate, as well as continued engagement in public-sector training and institutional development initiatives.

• Patrick Pérez – Chief Executive Officer at Kyutai

  • Speaker (2nd session)

  • Talk title: Entering the era of world models and physical AI

  • Abstract: TBD

  • Bio: Patrick Pérez is CEO at Kyutai, a non-profit open-science AI lab, based in Paris. Prior to this, Patrick was at Valeo as VP of AI and Scientific Director of valeo.ai (2018-2023), and with Technicolor (2009-2018), Inria (1993-2000, 2004-2009) and Microsoft Research Cambridge (2000-2004) as research scientist. His research interests lie in reliable multimodal AI for the benefit of all.

• Lionel Basdevant – Senior Product Director at Criteo

  • Speaker (2nd session)

  • Talk title: Agents for e-commerce

  • Abstract: Online shopping is ongoing a massive change, with AI agents ready to become your personal shopping assistants. But fulfilling the promise of agentic shopping is a challenge for brands, retailers, marketplaces, search engines, and AI companies alike.
    Criteo is leading the shift toward agentic commerce, by developing an Agentic Commerce Product Recommendation service designed to better serve both our clients and the shopper journey. This service is powered by Criteo’s commerce data—a unique traffic signal that differentiates Criteo and delivers deep reco that result in truly delightful shopping experiences.
    In this talk, we will give an overview of the ongoing eCommerce market change, the service Criteo is building, and explain the tech, data, and AI used to go beyond simple catalog browsing.

  • Bio: Lionel Basdevant is Senior Product Director at Criteo and leads the company’s efforts around Agentic AI. As AI transforms the web, Lionel is focused on building solutions that work at scale and fit the needs of all players in the ecosystem, first and foremost advertisers and media owners.

    Lionel has over 15 years of experience in digital advertising, product development, and project management. Prior to joining Criteo in 2012, he held project manager and consultant positions with InovenAltenor, Calyon Americas, and Capgemini. Lionel holds an engineering degree from the Ecole Centrale de Paris.

• Alberto Lumbreras – Staff Researcher at Criteo

  • Speaker (2nd session)

  • Talk title: Agents for e-commerce

  • Abstract: Online shopping is ongoing a massive change, with AI agents ready to become your personal shopping assistants. But fulfilling the promise of agentic shopping is a challenge for brands, retailers, marketplaces, search engines, and AI companies alike.
    Criteo is leading the shift toward agentic commerce, by developing an Agentic Commerce Product Recommendation service designed to better serve both our clients and the shopper journey. This service is powered by Criteo’s commerce data—a unique traffic signal that differentiates Criteo and delivers deep reco that result in truly delightful shopping experiences.
    In this talk, we will give an overview of the ongoing eCommerce market change, the service Criteo is building, and explain the tech, data, and AI used to go beyond simple catalog browsing.

  • Bio: Alberto is a researcher at the Criteo AI Lab and Scientific Advisor for Criteo’s Agentic Program. He leads research on LLM-based AI agents and their application to commerce, working at the intersection of AI research and product development. He helps product teams identify opportunities and translate them into production-grade agent solutions—from early prototypes to the systems powering Criteo’s agentic transformation.

• Patrick Gallinari – Professor at Sorbonne Université & Distinguished Researcher at Criteo

  • Speaker (3rd session)

  • Talk title: Generative foundation models for search and recommendation

  • Abstract: TBD

  • Bio: Patrick Gallinari is a professor at Sorbonne Université, affiliated with the Institute of Intelligent Systems and Robotics (ISIR) laboratory, and a distinguished researcher at Criteo AI Lab in Paris. A pioneer in the field of neural networks, his research focuses on statistical learning and deep learning, with applications spanning semantic data processing and complex data analysis. Currently, his work centers on scientific AI also known as AI4Science with a focus on physics-aware machine learning where he has made seminal contributions, and on NLP topics. He also holds a national AI chair titled “Deep Learning for Physical Processes with Applications to Earth System Science” (2020 – 2026) and he is currently co-PI of a chair “Deep Learning for Science: Modeling Fluid Dynamics in Engineering and Climate Physics” (2025 – 2030).

    Patrick has led the Machine Learning team (MLIA) for several years. He served as the director of the Computer Science Laboratory at Sorbonne University (LIP6) from 2005 to 2013, following six years as its deputy director. Additionally, he was the vice director of the Scientific Committee of the Faculty of Engineering at Sorbonne Université (formerly Université Pierre et Marie Curie) from 2010 to 2021.

• Maria Teleńczuk – Senior Researcher at Owkin

  • Speaker (3rd session)

  • Talk title: Stop cleaning data, focus on science: Multi-Agent System for End-to-End Biological Data Processing

  • Abstract: TBD

  • Bio: Maria Teleńczuk, PhD, is a senior research scientist at Owkin specializing in machine learning. Her work focuses on building ML systems for complex, high-dimensional and temporal biomedical data. More recently, she has been working with agent-based systems, including adapting and fine-tuning large language models for scientific workflows. She also contributes to open-source projects and support initiatives that promote inclusion in technology and science.

• Grégoire Mialon – Staff Research Scientist at Meta Superintelligence Lab

  • Speaker (3rd session)

  • Talk title: Evaluating frontier agents

  • Abstract: There is no progress in AI without good evaluations. However, as models become more and more capable, it becomes more and more difficult to create evaluations that are challenging for the best models while being rooted in actual use cases. This talk will cover evaluations for frontier AI agents, some of those we created, and their impact on the field.

  • Bio: Grégoire is a Staff Research Scientist at Meta Superintelligence Labs in Paris, working on LLM post-training and Agents. Before that, he was a post-doc in Yann LeCun’s team at Meta, FAIR. He got a PhD from Inria Paris and Grenoble.

• Anne Bouverot – France’s Special Envoy for AI, Chairperson of Ecole Normale Supérieure, Co-Chair of France’s AI & Digital Council

  • Fireside chat

  • Bio: Anne Bouverot is France’s Special Envoy for Artificial Intelligence and Chair of the Board of Directors at the École Normale Supérieure. After earning a PhD in computer science, she spent the majority of her career in the technology sector and now advises companies and scale-ups. Alongside economist Philippe Aghion, she co-chaired the report “AI: Our Ambition for France” in March 2024. As the President’s Special Envoy, she led the preparations for the AI Action Summit held in Paris in February 2025. She spent the first 20 years of her career at Orange, then became Director General of the GSMA, the global association of telecommunications operators, and later CEO of the digital security company Morpho.

Event organized by Criteo, with (in alphabetical order):

• Marieke Buis, Senior PR & Communication Manager France;

• Joaquín Engelmo Moriche, Staff Developer Relations Engineer;

• Jean-Yves Franceschi, Staff Researcher;

• Nathalie Laneret, VP Government Affairs & Public Policy;

• Dainora Magnier, R&D Support Coordinator;

• Mariia Vladimirova, Senior Researcher.